A suggestion

[Jerry Brunner]

Hi, this is Jerry Brunner, a faculty member in Statistics at the
University of Toronto. I just found out about PSPP, and I think you are
doing an extremely useful thing. I saw your request for help with
statistical code (back-end stuff). May I make a suggestion?

Why not treat PSPP as a front-end to R? The code is already written, it's
open source and GNU-protected, and the core is of stellar quality. Many of
the contributed packages (there are over 700, I believe) are completely
sound too, though of course I can't vouch for all of them.

The PSPP front end would be EXTREMELY valuable. Many users can't tolerate
the R syntax; they want a traditional stat package like SAS or SPSS.

And I bet all you'd need to implement a procedure would be to have
somebody tell you how to get the numbers you want (that is, the
numbers SPSS produces) using R. Lots of us could do that.  

Good luck and thank you for what you are doing,

Jerry